Spring-Cloud-Sleuth | Spring Cloud Sleuth Zipkin

You have configured the wrong option on the driver. METADATA_SCHEMA_REQUEST_TIMEOUT is the timeout for the requests to retrieve the schema.

The default request timeout for the Java driver is basic.request.timeout (see reference configuration):

To me, it looks like span and transaction names are missing (I only see HTTP GET instead of a name)

No, they are not missing. The reason you are seeing name as HTTP GET is due to preference for less cardinal names and the semantic conventions for tracing data. There is a detailed explanation about the naming conventions for the HTTP spans here https://github.com/open-telemetry/opentelemetry-specification/blob/main/specification/trace/semantic_conventions/http.md#name. Any data generated by auto-instrumentation libraries will adhere to the semantic conventions specification. I am guessing the visualisation from the linked vendor is coming from manual instrumentation where you as an end user can give any name (although it is recommended everyone uses less cardinal values but there is not enforcement there). I don't think there is anything you can "fix" here.

I am having a hard time understanding. Does it mean from 3.0/2022.0 onward, there is no Sleuth anymore? We need to use Micrometer instead?

Yes

How to keep existing features offered by Sleuth please?

We will be providing support for those via Micrometer & Micrometer Tracing.

does it means we can just remove the Sleuth dependency from the pom file, and instead add/replace with the micrometer one, and things will be the same, still be working?

If you don't have any custom Sleuth configuration then there should be nothing that you should do for observability to work. If you had custom Sleuth configuration you will need to port it to Micrometer Tracing which essentially is a copy of Sleuth so that should be mostly a package change. If you were doing Brave customizations then most likely you'll have nothing to change.

The error means that TransactionSynchronizationFactory must be declared as a bean and therefore its BeanFactory requirements would be fulfilled.

Please, revise all your volatile objects and consider to make some of them as global singleton beans.

I believe all of these could go to the main configuration, not dynamic: PollerMetadata, TransactionInterceptor, TransactionManager and TransactionSynchronizationFactory. The dynamic part in your solution is really just that FileReadingMessageSource.

It is better to use as less dynamic as possible for better performance and proper reusability. You also would avoid duplications whenever they are not needed.

I would even move that .handle(message -> System.out.println(message)) to the separate global singleton IntegrationFlow and use only a channel reference from this one to avoid extra duplication.

1. I created a new trace using newtrace method of tracer which returns a span.
2. Started this span.
3. Then added this span in the currentscope using withspaninscope method.
4. Started scoped span for the runnable task and finished both of these spans.

It worked.

Spring Cloud AWS is now a community project, see: https://awspring.io/ The current release train of Spring Cloud (2020.x), where you can find Sleuth 3.x does not include/support it.

You should use io.awspring.cloud:spring-cloud-aws-messaging:2.3.x instead of org.springframework.cloud:spring-cloud-aws-messaging:2.2.x with the current release train of Spring Cloud (2020.x).

Because of these, Sleuth does not instrument Spring AWS Messaging directly but if you can use SQS in a way that integrates with Spring Integration/Messaging/Stream/something that is already instrumented, this could work; i.e.: I would try using QueueMessageChannel and the annotations, I don't think QueueMessagingTemplate will work.

Another thing you can do is manually instrument it and propagate tracing information in the headers, see: TraceSpringMessagingAutoConfiguration

Check the compatibility matrix, you should use Spring Cloud 2020.0.x aka Ilford (Sleuth 3.x) with Spring Boot 2.4 (or 2.5).

Sleuth 2.2.8 is Spring Cloud Hoxton, it is not compatible with Boot 2.4.

Spring Cloud Sleuth pushes the sampling decision down to the tracer implementation, you need to create either a SamplerFunction or a Sampler to do this, please see the docs:

• Sleuth: Sampling
• Sleuth: Brave Sampling
• Brave: Sampling

Update: Sampling is an early decision because tracing has some overhead and early sampling can save your app from that. If you want to filter out your spans at the end, you can setup an always sampler and a SpanHandler where you can access the Span (error, tags, etc.) and you can also keep/drop them based on this (you can inject a sampler to your handler and apply its logic to the rest of your spans).

Please upgrade the project to the latest version of Spring Boot and Spring Cloud. Also use the BOM as the source of versions - do not set them manually. Please do not set the default properties manually cause as defaults they are already set.